Paid Ad Creative Refresh Workflow: What to Keep, Change, and Test Next

“We need fresh creative” can turn into a pile of new files without a clear reason for any of them.

The better starting point is to look at what changed. Is the same audience seeing the ad too often? Is attention falling while the landing page still works? Is the campaign producing conversions that do not become useful business outcomes? Or did the offer simply change?

I would use those answers to decide what the next request needs to do. Sometimes the right move is a new concept. Sometimes it is a focused variation. Sometimes the current ad should stay exactly where it is.

This paid ad creative refresh workflow separates that decision from the production work. The goal is a smaller, more useful queue, not more ads for the sake of activity.

A refresh date is a review point, not an expiration date

A schedule is still helpful. LinkedIn says click-through rates can decline when the same ad runs week after week and recommends refreshing ad copy or images at least monthly. It also recommends running two to four ad variations at a time. LinkedIn’s ad best-practice guidance

I would treat that as a useful reminder to review the account, not a rule that every ad must be replaced on day 30.

An ad that is still producing useful outcomes can remain as a control. A new creative can enter beside it. That keeps the account from losing a working direction while the new idea is still finding its footing.

Google gives similar guidance for Demand Gen campaigns: add new assets before removing old ones, keep top-performing assets during gradual evergreen refreshes, and allow enough time for new assets to ramp up. Google’s Demand Gen creative-refresh guidance

The exact timing will vary by platform, spend, audience size, and sales cycle. The principle is more durable: review on a schedule, then make the production decision from the evidence you actually have.

Start with the signal, then find the layer that needs work

One performance number rarely tells you what to design next.

A lower click-through rate can point to a tired visual or message. It can also reflect a different audience mix, more competitive inventory, or a delivery change. A higher cost per conversion could begin with the ad, the landing page, the form, the offer, or lead quality after the form.

I would review the campaign in four layers:

1. Exposure: reach, frequency, delivery, audience size, and where the ad appeared.
2. Attention: click-through rate, engagement, video behavior where relevant, and the response to each creative.
3. Post-click action: landing-page visits, form starts, conversions, and the connection between the promise in the ad and the next page.
4. Business outcome: qualified leads, orders, revenue, sales activity, or another result outside the ad platform.

This keeps a creative refresh from becoming the automatic answer to every campaign problem.

If attention is falling but post-click performance is stable, a new hook or visual direction may be a good next test. If attention is stable but the landing-page conversion rate drops, I would look at the page and offer before asking for ten more ads. If platform conversions look fine but lead quality is weak, the message may be attracting the wrong people or the conversion event may be too early.

Sort every ad into four production decisions

Once the review is complete, place each active creative into one of four states.

Google’s asset reporting uses labels such as Low, Good, and Best to help advertisers decide what to remove, improve, or build on. Google also says to compare assets with other assets of the same type. Google Ads asset reporting guidance

Those labels can inform the classification, but I would not let a platform label make the whole decision. Check delivery volume and the business outcome too. An attractive click is not automatically a useful lead, and a low-volume asset may still be in the learning stage.

Choose a concept, variation, or adaptation

These are three different production requests.

• New concept: a different reason to pay attention or act. The visual idea and message may both change.
• Variation: a controlled change inside an existing concept, such as a new headline or product image.
• Adaptation: an approved concept reworked for a different placement, aspect ratio, or viewing context.

If the current concept is still working, I would avoid rebuilding everything at once. Test a couple of versions and keep the control. If the concept has stopped answering the campaign question, small color and button changes will not create a genuinely new idea.

The earlier guides to building distinct static-ad concepts and resizing versus adapting ads go deeper on those decisions.

A hypothetical refresh decision

The important part is that the production request follows the diagnosis. The team is not changing the offer, landing page, audience, message, and layout all at once. It is making a deliberate creative change while protecting a useful reference point.

Copy this creative refresh request

Once the decision is clear, the request can be short. If the campaign goal, audience, or offer has changed substantially, use the fuller paid ad creative brief template instead.

PAID AD CREATIVE REFRESH REQUEST

Campaign and placements:
[Campaign, platform, ad format, placements, and current links]

Observed signal:
[What changed? Include the date range and enough context to read it.]

Business check:
[What happened after the platform conversion?]

Decision:
[Keep, iterate, replace, or watch]

What stays the same:
[Audience, offer, landing page, control ad, brand elements]

What changes:
[Concept, headline, image, proof point, layout, or adaptation]

Test question:
[What are we trying to learn?]

Deliverables:
[Number of concepts, versions, sizes, file types, and priority order]

Approved inputs:
[Copy, claims, product images, brand assets, and references]

Decision rule and review date:
[What result will inform the next action, and when will we review it?]

I would keep one active refresh request at a time. Get the first concept right, resolve the useful feedback, then move into the agreed versions and placement adaptations. That is usually faster than asking for every possible file before the direction is settled.

Name the files so the result can inform the next request

The feedback loop breaks when the media team cannot match the launched ad to the design file.

Use a naming pattern that records the concept, version, format, and date. For example:

product-proof_v2_square_2026-10-30.png
product-proof_v2_landscape_2026-10-30.png

Then keep the same identifier in the platform’s ad name where possible. At the next review, you should be able to connect the campaign result to the exact creative decision without opening ten nearly identical files.

Define the next decision before the new ads launch

A refresh is not finished when the exports are delivered.

Write down when the new work will be reviewed, what result matters, and what will happen next. The platform needs enough delivery to produce a useful signal, and the business may need more time to evaluate lead quality or revenue.

I would avoid declaring a winner from a few early clicks. Google notes that performance labels may remain in Learning until an asset has enough traffic, and that an Unrated asset may not have enough information yet. Google Ads asset reporting guidance

At the review, ask:

• Did the new creative earn more useful attention?
• Did that attention continue into the next action?
• Did the conversions become qualified leads, orders, or another useful outcome?
• Which part of the result is clear, and what still needs another test?

Then put each ad back into keep, iterate, replace, or watch. That makes the refresh a repeatable operating system instead of a one-time scramble for new graphics.

Sources

• LinkedIn Ads: Best practices for LinkedIn Ads
• Google Ads Help: About the asset report
• Google Ads Help: Refresh Demand Gen campaign creative
• Meta Blueprint: Creative diversification

Platform guidance checked October 30, 2026. Review current requirements and account-specific delivery before production.